Claude Scope: Inspect your UI, prompt your AI

Claude Scope records your browser tab, maps every component via accessibility snapshots, and writes your AI coding agent prompt automatically.

Claude Scope is a developer tool that turns your screen recording into a structured system prompt ready for Claude Code, Codex, or Cursor. Record your browser tab, let the dual-lane analysis engine inspect your UI, and copy a complete prompt — no manual screenshots or descriptions needed.

Quick Start

Sign up, connect your API key, and create your first recording in minutes.

How It Works

Learn how Claude Scope extracts frames, inspects accessibility trees, and synthesizes prompts.

Recording Guide

Set up a new recording, choose your target URL, and control when to stop.

API Reference

Integrate Claude Scope into your own workflow using the REST API.

How it works in three steps

1. Record your screen

   Open Claude Scope, enter a title and the URL you want to inspect, and click Start Recording. Interact with your app naturally — Claude Scope uses SSIM-based frame differencing to capture only meaningful UI state changes.

2. Dual-lane analysis

   After you stop recording, two analysis lanes run in parallel: a Vision AI lane that examines each extracted frame for UI elements, and a Playwright lane that captures a full ARIA accessibility snapshot of your seed URL.

3. Copy your prompt

   Claude Scope merges the visual timeline and structural inspection into a single, ready-to-paste system prompt. Choose your target agent — Claude Code, Codex, Cursor, or Raw — and copy it directly into your coding session.

Claude Scope is currently free during beta. You bring your own Anthropic API key for vision analysis. See Model Access to configure yours.